Embodiments of the present invention relate to anatomy segmentation, and more specifically, to computationally efficient anatomy segmentation through low-resolution multi-atlas label fusion and corrective learning.